The Impact of Vehicle Automation on the Safety of Vulnerable Road Users (Pedestrians and Bicyclists)

Autonomous systems are capable of transforming society with their ability to reduce risks and costs and improve efficiency. However, there are several challenges associated with the deployment of such autonomous systems largely stemming from the unknown parameters. This project is specifically interested in addressing vehicle automation. Semi-automation or partial automation has been existent for years starting with the installation of intelligent transportation systems (such as lane assist systems, collision warning systems and braking stability control systems) in vehicles. Literature notes the benefits of partial automation as well as indicates the costs associated with such automation primarily arising from lapses on the side of the human operator (driver), indicating subtle limits on the smooth transfer of control (TOC) between the human and the automation. Considering that full and extensive vehicle automation is almost a reality, it is critical to address the potential safety trickle down effects springing from automation (or levels of automation) on the safety of vulnerable road users like pedestrians and bicyclists. A pre-emptive look at how the TOC parameters associated with the engagement and disengagement of such systems affects the drivers' ability to detect and respond to vulnerable road users is warranted. As a precursor, this project is currently engaged in the process of evaluating and identifying the critical duration that governs seamless transfer of control. Specifically, the primary parameter being identified is the minimum TOC alerting time which is the minimum time beyond which the user has to take over manual control or re-engage. In the proposed project, there is interest in understanding the limits on TOC under varying levels/conditions of autonomy when the latent hazard is a potential vulnerable road user (pedestrian/bicyclist). Interest lies in not only determining how the driver responds to a vulnerable user threat under conditions of automation, additionally the project also seeks to understand whether the observed differences in scanning and yielding behavior is a function of the level of automation. It would be crucial to initially determine the optimal latent hazard headway time which is the time between the issue of command that TOC will take place in 'x' seconds and the location of the threat (vulnerable road user). The latent hazard headway time may vary as a function of the type of processing, the level of automation in the environment and the time the operator has been outside the loop. It would also appear that the yielding and scanning behavior of drivers (which is critical to vulnerable user safety) would be impacted by the optimal latent hazard headway time, level of automation and the time the operator has been outside the loop. It would be intuitive to hypothesize that the vulnerability of a pedestrian or bicyclist increases, as the latent hazard headway time decreases and the level of automation increases. However, it remains to be seen if when absolved of the continuous effort of maintaining lane position and velocity, can the driver pay enough attention to the roadway to successfully take over control in a pedestrian or bicyclist impedance event that occurs 'y' seconds after the TOC command has been issued 'x' seconds in advance, where 'y' > 'x' to ensure that the hazard headway time is not lesser 2 than the alerting time ('x' > 'y' leads to a potential situation where TOC from automation to user will not occur as it is deemed unsafe). The proposed study will be conducted on a driving simulator making it possible to measure eye behaviors like glance fixations and vehicle responses (velocity, lateral lane deviation) occurring in a standard simulator environment under a range of conditions. The overall objective and contribution of this research is threefold: (1) to identify the optimal latent hazard headway time in an autonomous system when the automation is engaged; (2) given the optimal latent hazard headway time, to evaluate drivers' scanning and yielding behavior towards vulnerable road users in a driving simulator environment; and (3) to document and disseminate the research findings in conferences and journals.

  • Supplemental Notes:
    • Research project done at University of Massachusetts-Amherst.


  • English


  • Status: Active
  • Contract Numbers:


  • Sponsor Organizations:

    Research and Innovative Technology Administration

    University Transportation Centers Program
    1200 New Jersey Avenue
    Washington, DC  United States  20590
  • Project Managers:

    Kline, Robin

  • Performing Organizations:

    Safety Research Using Simulation University Transportation Center (SaferSim)

    University of Iowa
    Iowa City, IA  United States  52242
  • Principal Investigators:

    Fisher, Donald

    Samuel, Siby

  • Start Date: 20150401
  • Expected Completion Date: 20170630
  • Actual Completion Date: 0
  • Source Data: RiP Project 39212

Subject/Index Terms

Filing Info

  • Accession Number: 01556671
  • Record Type: Research project
  • Source Agency: Safety Research Using Simulation University Transportation Center (SaferSim)
  • Contract Numbers: DTRT13-G-UTC53
  • Files: UTC, RiP
  • Created Date: Mar 12 2015 1:00AM